Relayed signaling between mesenchymal progenitors and muscle stem cells ensures adaptive stem cell response to increased mechanical load

Adaptation to mechanical load, leading to enhanced force and power output, is a characteristic feature of skeletal muscle. Formation of new myonuclei required for efficient muscle hypertrophy relies on prior activation and proliferation of muscle stem cells (MuSCs). However, the mechanisms controlling MuSC expansion under conditions of increased load are not fully understood. Here we demonstrate that interstitial mesenchymal progenitors respond to mechanical load and stimulate MuSC proliferation in a surgical mouse model of increased muscle load. Mechanistically, transcriptional activation of Yes-associated protein 1 (Yap1)/transcriptional coactivator with PDZ-binding motif (Taz) in mesenchymal progenitors results in local production of thrombospondin-1 (Thbs1), which, in turn, drives MuSC proliferation through CD47 signaling. Under homeostatic conditions, however, CD47 signaling is insufficient to promote MuSC proliferation and instead depends on prior downregulation of the Calcitonin receptor. Our results suggest that relayed signaling between mesenchymal progenitors and MuSCs through a Yap1/Taz-Thbs1-CD47 pathway is critical to establish the supply of MuSCs during muscle hypertrophy.


Yap/Taz nuclear translocation
Skeletal muscle is an essential organ for locomotion, metabolism, and life activities. Its plasticity is reflected by an unparalleled ability to adapt to external and internal physiological alterations. For example, enhanced mechanical load induces skeletal muscle remodeling and, consequently, increasing mus-cle mass and strength, known as muscle hypertrophy. Hypertrophy goes along with enhanced protein synthesis and an increase in myonuclei (Bamman et al., 2018), with the latter being completely dependent on muscle stem (satellite) cells (MuSCs) (Egner et al., 2016;McCarthy et al., 2011). Although a few studies suggested that MuSCs are less important during the early phase of hypertrophy (Murach et al., 2017), the majority of studies indicate an essential role of MuSCs for efficient muscle hypertrophy, either by myonuclear accretion or as a source of paracrine factors (Egner et al., 2016;Fry et al., 2017;Fukuda et al., 2019;Goh and Millay, 2017;Moriya and Miyazaki, 2018). Although the molecular mechanisms regulating MuSC behavior are critical for muscle hypertrophy, they remain poorly investigated. Because MuSCs are located on contracting and relaxing myofibers, it has been assumed that MuSCs sense changes in myofiber activity. However, MuSCs seem to ignore the basal activity of skeletal muscle and remain in a quiescent state. MuSCs are activated and expand only when skeletal muscle is exposed to a strong mechanical stimulus. At present, it is unknown how increased mechanical load switches MuSCs from quiescence to activation. Myofiber injuries might be one possible trigger for expansion of MuSCs under increased mechanical load (Ebbeling and Clarkson, 1989). However, massive damage or even death of myofibers, required for massive expansion of MuSCs, does not occur in overloaded muscles (Darr and Schultz, 1987;Fukada et al., 2020). Other mechanisms, such as edema formation (swelling caused by excess fluid), responsible for early weight gain of muscles during resistance training, and microfractures of the sarcolemma are more likely signals (Damas et al., 2018;Fukada et al., 2020). In addition, myofiber-derived factors that are released during enhanced mechanical load may contribute to expansion of MuSCs (Guerci et al., 2012;Serrano et al., 2008). Nevertheless, and despite much work on the relationship between mechanical loading and MuSC proliferation, our understanding of these processes is immature, and research questions remain.
It has been assumed that myofibers are primarily responsive for sensing enhanced mechanical load in skeletal muscle, although interstitial mesenchymal cells might be involved as well. Mesenchymal progenitors (Uezumi et al., 2010), also known as fibro/adipogenic progenitors (FAPs) , are characterized by expression of platelet-derived growth factor receptor alpha (Pdgfra) and are responsible for pathological fibrosis and fat accumulation in skeletal muscle Uezumi et al., 2010Uezumi et al., , 2011. On the other hand, mesenchymal progenitors are also involved in homeostasis and regeneration of skeletal muscle Lemos et al., 2015;Uezumi et al., 2021;Wosczyna et al., 2019). However, knowledge about the role of mesenchymal progenitors in regulation of muscle hypertrophy is lacking.
Yes-associated protein 1 (Yap1) and transcriptional coactivator with PDZ-binding motif (Taz, also known as Wwtr1), are wellknown effectors of the Hippo kinase cascade, required for proliferation, survival, and tumorigenesis in many cell types (Yu et al., 2015). Yap1/Taz also function as sensors and transducers of mechanical signals (Dupont et al., 2011), which might important for induction of muscle hypertrophy by mechanical load. In fact, overexpression of YAP induces muscle hypertrophy (Goodman et al., 2015;Watt et al., 2015). However, total Yap1 and phosphorylated Yap1 levels peak only 7 days after onset of surgical overload in FVB/N mice, much later compared with mTORC1 signaling, which peaks 2 days after induction of overload (Goodman et al., 2015). We found previously that Yap1 accumulates in nuclei of calcitonin receptor (CalcR) mutant MuSCs and demonstrated that the CalcR-protein kinase A axis suppresses Yap1 activity in quiescent MuSCs .
CalcR mutant and constitutively active Yap1-expressing MuSCs (Tremblay et al., 2014) did not exhibit substantial cell division as in overloaded muscle. These results suggest that Yap1 does not have a primary function in myofibers and MuSCs, arguing for an alternative mechanism that is required to induce MuSC expansion in response to increased mechanical load.
Here we investigated the role of mesenchymal progenitors for proliferation of MuSC in overloaded muscle. We further analyzed the role of Yap1/Taz in these cells and assessed their effect on MuSC expansion during increased mechanical load.

RESULTS
Mesenchymal progenitors are essential for myonuclear accretion from muscle stem cells in overloaded muscle We speculated that Pdgfra + mesenchymal progenitors (hereafter Pa + cells) have a previously underappreciated function in muscle hypertrophy. Therefore, we studied Pa + cells in overloaded muscles in which the distal tendons of the gastrocnemius and soleus muscles were cut (tenotomy) to induce compensatory hypertrophy in the plantaris (PLA) muscle with little degenerative damage to myofibers . Proliferating Pa + cells on days 2, 4, and 7 after tenotomy were counted after labeling with 5ethynyl-2 0 -deoxyuridine (EdU) or staining against an activation/ proliferation marker, Ki67, in Pdgfra-H2B-EGFP (Pa-EGFP) or C57BL/6 mice ( Figure 1A). Two and 4 days after tenotomy, approximately 2% and 13% of Pa + cells were positive for EdU, and 5% and 30% of Pa + cells were positive for Ki67, respectively ( Figures 1B, 1C, S1A, and S1B). Only a few Pa + cells were labeled with EdU 6-7 days after tenotomy, and approximately 3% of Pa + cells expressed Ki67 on day 7 after tenotomy ( Figures  1B and 1C). The number of Pa + cells was increased slightly on day 4 (1.6-fold) and 7 days (2-fold) after tenotomy in Pa-EGFP ( Figure 1D). The same increase was also detected in C57BL/6 mice 7 days after tenotomy ( Figure S1C). In addition, similar to MuSCs, the cell size and granularity of Pa + cells in fluorescence-activated cell sorting (FACS) analyses were increased 2 days after tenotomy, ( Figures 1E, 1F, S1D, and S1E). During this period, apoptotic Pa + cells were not detected ( Figure S1F), indicating that Pa + cells are activated in response to increased mechanical load and show a transient and moderate increase in proliferation.
To determine the role of activated Pa + cells in overloaded muscles, we depleted Pa + cells by treating Pdgfra CreERT :: Rosa DTA (Pa-DTA) mice with tamoxifen and confirmed successful depletion of Pa + cells ( Figure S2A), as reported previously (Uezumi et al., 2021). Next, PLA muscle hypertrophy was induced in Pa + cell-deficient mice by tenotomy, followed by EdU injections (Figures 2A). Although the sham muscle weight was decreased by Pa + cell depletion, correlating with our previous study ( Figure S2B), the increased ratio of overloaded muscle weight to sham muscle weight was similar in control (Pdgfra CreERT/+ ) and Pa-DTA mice 7 days after tenotomy (Figure 2B), suggesting that early responses, including edema, were similar in control and Pa-DTA mice. However, ablation of Pa + cells suppressed the increase in total myonuclei after tenotomy ( Figure 2C). Likewise, newly formed EdU + myonuclei were reduced markedly in Pa-DTA compared with control mice (Figures 2D and 2E). M-cadherin (M-cad) + and M-cad + EdU + cells in loaded muscles were also reduced in Pa-DTA mice (Figures 2F and 2G). Because approximately 20% of Pa + cells survive in Pa-DTA mice ( Figure S2A), we assume that the remaining Pa + cells are sufficient to promote accretion of some MuSC-derived EdU + nuclei into myofibers.
To further investigate the effects caused by loss of Pa + cells on MuSCs in overloaded muscles, single myofibers were isolated from overloaded PLA muscle, and Pax7 + cells numbers were counted on day 4, 7, and 10 after tenotomy ( Figure 2H). Consistent with the results of myonuclear accretion, the numbers of Pax7 + cells in the overloaded muscle of Pa-DTA mice were reduced remarkably compared with the control 4-10 days after tenotomy ( Figure 2I). To elucidate the role of Pa + cells in MuSC activation, we compared the expression of Ki67 and MyoD between control and Pa-DTA mice. Our previous study indicated that, upon MuSC activation, Ki67 expression occurs 2-4 days after tenotomy and that 60%-80% of MuSCs are Ki67 + on day 4. In addition, the majority of MuSCs proliferate without apparent expression of MyoD . Therefore, we used expression of Ki67, but not MyoD, as a marker for MuSC In all Figures excluding Figure 5G, one symbol means the result of one mouse. All data are presented as mean ± SD. N.S., not significant. Nuclei were counterstained with DAPI. See also Figure S1. Article activation 4 days after tenotomy. Notably, Ki67 + Pax7 + numbers were reduced markedly in Pa-DTA mice on day 4 after tenotomy (Figures 2J and 2K). Although the frequency of MyoD + or myogenin + cells per Pax7 + or M-cad + cell in Pa-DTA mice was comparable with those in the control ( Figures 2K and 2L), suggesting that loss of Pa + cells did not influence myogenic differentiation of overloaded MuSCs. Next we investigated the effect of Pa + cell ablation 11-12 weeks after synergistic ablation (SA), at a late stage of muscle hypertrophy. Although the relative increases in PLA muscle weight and myofiber size in Pa-DTA mice 2 weeks after tenotomy were comparable with control mice ( Figures S2C-S2E), these increases were blunted markedly 11-12 weeks after SA ( Figures  S2F-S2H). Importantly, blunted myonuclear accretion was confirmed even within this period ( Figure S2I). These results suggest that Pa + cells are essential for efficient muscle hypertrophy, including regulation of MuSC expansion.

Yap1 target gene expression in overloaded mesenchymal progenitors
To elucidate the mechanism by which Pa + cells affect MuSCs in overloaded muscles, we analyzed the transcriptomes of purified Pa + cells in sham or overloaded muscles on day 2 after tenotomy ( Figure 3A). Volcano plot analyses revealed 904 and 638 upregulated and downregulated genes, respectively, in Pa + cells from overloaded muscles ( Figure 3B). Gene Ontology (GO) analyses suggested activation of Pa + cells because cell cycle-related gene sets were upregulated in overloaded Pa + cells ( Figure S3A). To further investigate the cell cycle-related gene signature in overloaded Pa + cells, gene set enrichment analyses (GSEA) were performed using oncogenic signature gene sets from the Molecular Signature Database (https:// www.gsea-msigdb.org/gsea/). The gene set ''CORDENONSI YAP CONSERVED SIGNATURE'' was ranked as the top gene signature in this analysis (Figures 3C and 3D), as evidenced by upregulation of many prototypic Yap1 target genes in the overloaded Pa + cells ( Figure 3E), including Ctgf. Consistent with this finding, we observed nuclear localized Yap1, which is associated with transcriptional activity, in Pa + cells located at the distal half of the PLA muscle on days 1 and 2 after tenotomy ( Figures 3F and 3G). Nuclear localized Yap1 appeared to be linked to increased mechanical load because Yap1 was rarely observed in nuclei of corresponding regions in shamoperated PLA muscle ( Figures 3F and 3G). These results suggest that increased mechanical load leads to activation of Yap1 and induction of its downstream targets in Pa + cells.
To investigate the influence of Yap1/Taz in overloaded Pa + cells on MuSC behavior, M-cad + and EdU + M-cad + cells were counted in sham and overloaded muscles of Cont and cdKO mice. The results indicated a blunted response to overload in cdKO MuSCs compared with Cont MuSCs (Figures 4E and 4F). Next we investigated the effect of Yap1/Taz loss in overloaded Pa + cells on MuSC activation. In contrast to Pa-DTA mice, Ki67 expression in Pax7 + cells was not changed significantly in Cont, cdKO, or Y-cKO mice, whereas expansion of       Pax7 + cells was blunted in cdKO mice ( Figure 4G). These results indicate that Yap1/Taz activation in Pa + cells is critical for proliferation but not activation of MuSCs in overloaded muscle. Consistent with the results in Pa-DTA mice ( Figures 2K  and 2L), the frequency of MyoD + or myogenin + cells in Pax7 + or M-cad + cells did not change in cdKO mice ( Figures 4G and  4H), suggesting that Yap1/Taz in Pa + cells do not govern myogenic differentiation of MuSCs. We also examined the effect of Yap1/Taz loss in muscle hypertrophy 12 weeks after SA and determined that the increase in myofiber size and myonuclear number was blunted in cdKO as well as in Pa-DTA mice ( Figures  S4A-S4C), indicating that Yap1/Taz in Pa + cells are critical for substantial muscle hypertrophy.
To further elucidate the role of Yap1/Taz in Pa + cells, RNA sequencing (RNA-seq) analyses of Cont (Pdgfra CreERT/+ :: Yap1 +/+ ::Taz +/+ ) and cdKO Pa + cells were performed ( Figure 4I). The volcano plot analyses indicated that the increase in tenotomy-induced gene expression was attenuated in cdKO Pa + cells compared with Cont cells ( Figure 4J; 987 versus 291 genes). In addition to the expected decrease in Yap1 and Taz levels, Yap1 target genes were also reduced in cdKO Pa + cells compared with Cont from overloaded muscles (Figures 4K-4M). In addition, cell cycle-related GO terms, a signature of loaded Pa + cells, were not upregulated in cdKO Pa + cells (Figure S3B). These results indicate that increased mechanical load-induced gene expression and activation of Pa + cells strongly depend on Yap1/Taz.
Mesenchymal progenitor-derived Thrombospondin-1 promotes muscle stem cell proliferation Bioinformatics analysis of corresponding ligand-receptor pairs in Pa + cells and MuSCs led us to focus on Thbs1 and CD47. Thbs1, coding for Thrombospondin-1, was highly expressed in Pa + cells in response to overload ( Figure S3C), whereas its receptor, CD47, was detected in MuSCs 4 days after tenotomy in quiescent and proliferating MuSCs ( Figure S3D). Moreover, Thbs1 is a Yap1 target gene ( Figure 3E), as indicated by reduced expression in Yap1/Taz-deficient Pa + cells ( Figures 4L and 4M).
We corroborated that expression of Thbs1 increased specifically in Pa + cells 2 days after tenotomy but not in MuSCs and myofibers ( Figure 5A). RNA-seq analyses also indicated nonincreased expression of Thbs1 in loaded MuSCs, although other Yap1 target genes were increased ( Figure S3E). Furthermore, we determined Thbs1 protein levels in response to overload in muscles and blood by ELISA, which unveiled increased Thbs1 protein concentrations in PLA muscle but not in blood ( Figure 5B). Likewise, Pdgfra + Thbs1 + cells and Thbs1 + areas were frequently detected in overloaded PLA muscle of wild-type (WT) mice but not in sham and overloaded PLA muscle of Pa-DTA and cdKO mice ( Figures 5C, 5D, and S4D-S4F). The analyses of Thbs1 mRNA expression in the PLA muscle also suggested that Pa + cells were the main cell source in loaded muscles because Thbs1 mRNA was reduced in Pa-DTA and cdKO mice ( Figures  5D and S4G). In particular, we noticed that a part of Thbs1 was involved in the basal lamina surrounding myofibers and MuSCs, suggesting that Thbs1 may directly affect MuSC behavior in loaded muscles ( Figure 5E). These results indicate that Thbs1 is secreted by Pa + cells from overloaded muscles in a Yap1/ Taz-dependent manner and acts locally in muscles undergoing hypertrophy.
Thbs1 is a matricellular protein that can interact with a variety of molecules, including the extracellular matrix, cytokines, receptors, and proteases (Resovi et al., 2014). Thbs1-KO mice exhibited some abnormalities, including increased heart weight in males and a decline in muscle weight normalized to body weight in males and females, as reported previously ( Figure S5A; Malek and Olfert, 2009). Therefore, to investigate the effect of Thbs1 derived from Pa + cells on MuSCs, Pa + cells were isolated from WT or Thbs1-KO mice and cultured because nuclear localization of Yap1 and Thbs1 expression are detected in cultured Pa + cells ( Figure S5B). Importantly, supernatants from Thbs1-KO Pa + cells exhibited lower MuSC proliferation on isolated myofibers compared with supernatants from WT Pa + cells at 72 h (Figure 5F). Proliferation of Thbs1-KO Pa + cells was similar to that of WT Pa + cells ( Figure S5C). These results strongly indicate that Pa + cell-derived Thbs1 accelerates MuSCs proliferation in loaded muscles.
To further examine the importance of Thbs1-CD47 signaling in MuSCs, we generated MuSC-specific CD47 cKO mice (CD47-cKO; Pax7 CreERT2 ::Cd47 flox/flox ) ( Figure 6E). As expected, CD47 expression detected in Cont MuSC from sham or loaded muscles was absent in MuSCs from CD47-cKO muscles but not in other cell types, demonstrating efficient cell-type-specific deletion of CD47 ( Figures S5D and S5E). CD47-cKO mice revealed a remarkable reduction in EdU-labeled new myonuclei ( Figures  6F and 6G), and we observed fewer M-cad + and M-cad + EdU + cells in CD47-cKO than in Cont mice ( Figures 6H and 6I). Although increased Pax7 + cell numbers were not observed in Cont or CD47-cKO mice 4 days after tenotomy, loss of CD47 in MuSCs did not affect Ki67, MyoD, and myogenin expression ( Figures 6J and 6K). The non-increased number of MuSCs likely results from delayed MuSC proliferation in Pax7 CreERT2 mice possessing only one Pax7 allele. Muscle hypertrophy was blunted 12 weeks after SA but not over 7 days after tenotomy ( Figure S6). Further, the effect of PKHB1 disappeared in CD47-cKO mice ( Figure 6D), indicating that direct stimulation of CD47 signaling induces MuSC proliferation in loaded muscle. These findings indicate that the Thbs1-CD47 axis is critical for proliferation of MuSCs in overloaded muscle.

The CD47 agonist induces MuSC expansion in CalcR mutant mice
To further corroborate a critical role of CD47 signaling for MuSC expansion, C57BL/6 mice were treated with PKHB1, followed by assessment of MuSC numbers. Interestingly, PKHB1 treatment alone was insufficient to activate quiescent MuSCs and induce their expansion ( Figure 7A, 7B, S7A, and S7B), suggesting that additional factors are required for overload-mediated MuSCs proliferation.
Re-analysis of RNA-seq data of overloaded MuSCs  indicated decreased expression of the quiescence-specific genes Calcr, Tenm4, and Chrdl2 in MuSCs from overloaded muscles ( Figure S3D). We focused our further analysis on CalcR for the following reasons. First, CalcR signaling is an important pathway for maintaining the quiescent state in MuSCs. Although approximately 15%-20% of MuSCs express Ki67, they do not show substantial cell division in Calcr-cKO mice . Second, we observed decreased CalcR protein levels in MuSCs from overloaded muscles ( Figures 7C and 7D). Moreover, we found an inverse rela-tionship between CalcR and Ki67 expression in MuSCs from overloaded muscles ( Figure 7E). Similarly, CalcR expression was retained in Pa-DTA-derived MuSCs without an increase in Ki67 expression, whereas CalcR expression was decreased in cdKO MuSCs showing increased Ki67 expression (Figures 7F  and S7C). We therefore investigated whether downregulation of CalcR might enable expansion of MuSCs in response to PKHB1. We treated MuSC-specific CalcR mutant mice (Calcr-cKO) with PKHB1 and counted the number of MuSCs ( Figure 7G). Strikingly, PKHB1 administration increased the number of MuSCs in Calcr-cKO mice ( Figure 7H) and led to detection of increased numbers of new EdU + myonuclei and EdU + M-cad + cells in Calcr-cKO mice (Figures 7I and 7J). These results demonstrate that the coordinated regulation of Thbs1/CD47 and CalcR signaling observed in response to increased mechanical load is sufficient to induce MuSC proliferation.

DISCUSSION
In this study, we demonstrated an essential role of mesenchymal progenitors in regulating MuSC expansion during muscle hypertrophy. We show that increased mechanical load induces transcriptional activation of Yap1/Taz in mesenchymal progenitors, leading to expression of their target genes, including Thbs1. We establish that Thbs1 promotes MuSC proliferation via activation of its receptor CD47 and requires concomitant downregulation of CalcR. Our results identified a Yap1/Taz-Thbs1-CD47 signaling axis that enables mesenchymal progenitors to instruct MuSC expansion during mechanical overload. In addition, our study uncovers a conceptually unexpected paradigm: mesenchymal progenitors respond to mechanical cues and steer MuSC proliferation in overloaded muscle via paracrine Thbs1.
Thbs1 is a large, 150-kDa glycoprotein identified as a major secretory product of thrombin-stimulated platelets. Each domain of Thbs1 interacts with different molecules, resulting in multiple functions, including cell adhesion, motility, proliferation, and survival. The C-terminal domain of Thbs1 is necessary for binding to CD47. We used PKHB1, the first-described serumstable, soluble CD47-agonist peptide, to activate CD47 in MuSCs, and PKHB1 is a potential therapeutic tool against leukemia (Martinez-Torres et al., 2015). CD47 also functions as a ligand for SIRPa, and blockade of CD47-SIRPa interaction stimulates activation of the immune system, which is considered a potential therapeutic strategy against cancer (Chao et al., 2012). Because infiltration of macrophages, representative SIR-Pa-expressing cells, is limited in overloaded muscles, our results suggest that Thbs1 is the predominant ligand involved in activation of CD47 signaling in MuSCs. Further studies are required to determine the potential effect of PKHB1 on muscle regeneration by promoting MuSC proliferation in muscular dystrophies.
Consistent with our findings, it has been reported that Thbs1 expression is increased greatly in skeletal muscle following (F) Experimental scheme for analyzing the effect of Pa + cell-derived Thbs1 on Pax7 + cell expansion. Images show Pax7 + cells on freshly isolated myofibers (t = 0) or on myofibers cultured for 72 h with the supernatant of WT or Thbs1-KO Pa + cells. The graph indicates the average number of Pax7 + cells on myofibers. Pax7 + cell number was calculated by counting more than 14 myofibers per one mouse. Scale bar, 50 mm. (G) Effect of PKHB1 on myoblast proliferation. The y axis indicates the relative intensity of luminescence with respect to the beginning (2 h after seeding) of cultivation. **p < 0.01, PKHB1 30 mM versus Cont; #p < 0.05, PKHB1 10 mM versus Cont. See also Figures S4 and S5. active training in mice and humans (Hoier et al., 2013;Olfert et al., 2006). The functions of Thbs1 related to skeletal muscle biology, in particular activation of transforming growth factor b (TGF-b) and anti-angiogenesis, are relatively well investigated. Thbs1 is required for release of TGF-b activation from its latent form, which occurs independent of CD47, and promotes fibrosis in dystrophic muscles (Cohn et al., 2007). In line with this, functional expression of Col1a1, a target of TGF-b, increases in mesenchymal progenitors during overload, indicating additional functions of Thbs1 in skeletal muscle hypertrophy. Stimulation of TGF-b activity by Thbs1 also suggests that therapeutic application of Thbs1 needs to done with care to avoid excessive fibrosis. It has been also demonstrated that Thbs1 inhibits cGMP signaling by binding to CD36 and CD47, which suppresses angiogenesis. Increased capillary numbers in Thbs1-null skeletal muscle argue for a physiological role of Thbs1 in inhibition of angiogenesis (Malek and Olfert, 2009). However, the role of Thbs1 upregulation in exercise-triggered angiogenesis is not well understood (Hoier et al., 2013), although induction of angiogenesis by exercise training has been well documented (Olfert et al., 2006). Moreover, the cellular source of Thbs1 remains unidentified. Despite these limitations, our data reveal that mesenchymal progenitors are an important cellular source of Thbs1 and demonstrate an unexpected role of Thbs1 in expansion of MuSCs during muscle hypertrophy. The signaling processes downstream of CD47 in MuSCs are still unclear. CD47 acts via Gi-cyclic AMP (cAMP) signaling in several cell types (Frazier et al., 1999;Manna and Frazier, 2004;Yao et al., 2011), which corresponds to increased MuSC proliferation and differentiation upon constitutive activation of Gai2 (Q205L) (Minetti et al., 2014). In our study, however, neither recombinant Thbs1 nor PKHB1 reduced intracellular cAMP levels in forskolin-stimulated C2C12 cells (data not shown). In addition, PKHB1 alone did not induce exit from quiescence, which is secured by CalcR-cAMP-protein kinase A (PKA) signaling in MuSCs , suggesting that CD47 signaling in MuSCs does not prevent cAMP accumulation via Gi.
Localization of mesenchymal progenitors within the interstitium can be separated into three regions: endomysium, perimysium, and epimysium. Using single-cell RNA-seq analyses, Muhl et al. (2020) identified Thbs1 + Thbs4 + Pdgfra low and Thbs1 low Thbs4 + Pdgfra + cells in the perimysium and the interface between the perimysium and endomysium, respectively (Muhl et al., 2020). However, the majority of Pdgfra + cells in skeletal muscle do not express Thbs1 or Thbs4 or, if at all, then at low levels. Our own RNA-seq analyses in this study indicate that expression of Thbs1 and Thbs4 in mesenchymal progenitors is only low or negative in PLA muscle under baseline conditions but increases dramatically during overload. This finding is in line with immunostaining showing the presence of Thbs1 in the endomysium surrounding myofibers in overloaded muscles. Our results suggest that Pdgfra + cells in the endomysium start to express Thbs1 and Thbs4 in response to increased mechanical load.
Pdgfra is specific to mesenchymal progenitors in skeletal muscle, although Pdgfra + cells are present in several other tissues (Uezumi et al., 2010). Hence, ablation of skeletal muscleresident Pdgfra + cells or inactivation of Yap1/Taz were not restricted to skeletal muscle. Thus, we cannot completely exclude an influence of non-muscle Pdgfra + cells. However, we compared overloaded muscle with contralateral sham muscle in all experiments, confirming that effects on MuSCs depend on the local overload. In addition, the absence of any abnormality in cdKO mice under steady-state conditions and the unaltered levels of Thbs1 protein in the blood suggest that inactivation of Yap1/Taz or ablation of Pdgfra + cells in non-muscle tissues do not have a substantial effect on muscle physiology.
The discovery that MuSC can be expanded in non-injured and non-exercised muscles by CD47 signaling when CalcR expression is downregulated has important implications: First, it indicates that MuSC activation and proliferation during muscle hypertrophy do not require myofiber damage. Second, it provides a new approach to expand endogenous MuSCs in vivo for therapeutic purposes. Because MuSCs supply nuclei and mitochondria that are affected during aging, artificial expansion of MuSCs may have therapeutic potential for several disorders. Similarly, therapeutic expansion of MuSC might improve the chance of successful gene delivery into MuSCs for treatment of human muscle dystrophies.
Limitations of the study Yap1/Taz are known mechanosensors that respond to increased mechanical load and induce transcription of their target genes (Dupont et al., 2011). However, in addition to mechanical cues, alternative pathways might also be responsible for nuclear Figure 6. The Thbs1-CD47 axis is indispensable for MuSC proliferation in overloaded muscle (A) Experimental scheme for analyzing the effect of anti-CD47 inhibitory antibodies on myonuclear accretion in C57BL/6 mice. (B) Immunostaining of Dys and EdU in Ope 7 day muscle of C57BL/6 mice injected with Cont IgG or anti-CD47 antibody. The graph indicates the number of EdU + myonuclei in Sham and Ope muscles in the PBS (n = 6), Cont IgG (n = 6), or anti-CD47 antibody (n = 11) groups. (C) Experimental scheme for analyzing the effect of CD47 agonist (PKHB1) on myonuclear accretion in C57BL/6 or CD47-cKO mice. (D) Immunostaining of Dys and EdU in Ope 7 day muscle of C57BL/6 mice injected with PBS (Cont) or PKHB1. The graphs indicate the number of EdU + myonuclei in Sham and Ope muscles in Cont and PKHB1 groups of C57BL/6 (left; Cont, n = 9; PKHB1, n = 8) or CD47-cKO (right; Cont, n = 4, 2 M, 2 F; PKHB1, n = 5, 3 M, 2 F) mice. CD47-cKO mice were treated with tamoxifen 5 days before tenotomy. (E) Experimental scheme for analyzing the effect of MuSC-specific CD47 depletion on myonuclear accretion in overloaded PLA muscle. (F) Immunostaining of Dys and EdU in Ope 7 day muscle of Cont and CD47-cKO mice. (G) The number of EdU + myonuclei in Sham or Ope muscles from Cont (n = 8; 3 M, 5 F) or CD47-cKO (n = 10; 3 M, 7 F) mice. (H) Immunostaining of M-cad (red), LNa2 (green) and EdU in Ope 7 day muscle of Cont or CD47-cKO mice. (I) The number of M-cad + cells (left) or M-cad + EdU + cells per section (right) in Sham or Ope muscles of Cont (n = 8; 3 M, 5 F) or CD47-cKO (n = 10; 3 M, 7 F) mice. (J) Number of Pax7 + cells and frequency of Ki67 + and MyoD + cells in Pax7 + cells on a single myofiber of Sham or Ope 4 day muscle from Cont (n = 6; 3 M, 3 F for Pax7 + and Ki67 + ; n = 3; 1 M, 2 F for MyoD + ) and CD47-cKO (n = 6; 2 M, 4 F for Pax7 + and Ki67 + , n = 3; 2 M, 1 F for MyoD + ) mice. Pax7 + cell number and frequency were calculated by counting 14-30 myofibers per mouse. (K) The frequency of myog + cells in M-cad + cells from Ope muscles of Cont (n = 4; 3 M, 1 F) and CD47-cKO (n = 4; 3 M, 1 F) mice. Arrowheads indicate EdU + myonuclei or M-cad + cells. Scale bar, 50 mm. See also Figures S5 and S6. translocation of Yap1/Taz. It seems possible that mechanical forces lead to release of growth factors retained in the extracellular matrix, which then induce Yap1/Taz activation. In fact, mechanical overload-dependent release of hepatocyte growth factor (HGF) from the extracellular matrix has been reported (Tatsumi, 2010). Another possibility is that myofibers activate mesenchymal progenitors in response to overload; Reddy et al. (2020) demonstrated that myofiber-secreted succinate acts on mesenchymal progenitors upon exercise (Reddy et al., 2020). Considering the early nuclear translocation of Yap1 in overloaded mesenchymal progenitors, mechanical cue-dependent Yap1/Taz activation seems to be the most likely mechanism for mesenchymal progenitor-mediated MuSC proliferation in overloaded muscle. However, the lack of appropriate tools has limited our ability to prove that mechanical cues directly activate Yap1/Taz in mesenchymal progenitor cells in vivo.
Increasing evidence suggests that biological sex has an important effect on several biological processes, including regulation of muscle stem cells. Although we did not find obvious differences among different biological sexes, our study lacks the statistical power to make conclusive statements regarding potential effects of biological sex on the mesenchymal progenitor-Thbs1-CD47 axis in loaded muscles.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following: with the accession number GSE159082 and GSE162827. Gene expression data for myogenic cells at day 4 after tenotomy are available under accession number GSE135903 .

RT-PCR analysis
Total RNA was extracted from sorted cells by using Trizol LS and a QIAGEN RNeasy Micro Kit (QIAGEN) according to the manufacturer's instructions and then reverse-transcribed to cDNA by using a QuantiTect Reverse Transcription Kit (QIAGEN). Total RNA of plantaris muscle was extracted using miRNeasy Mini Kit (QIAGEN) after crushing by Shakeman (BMS, Bio medical science). Specific forward and reverse primers used for optimal amplification of the reverse-transcribed cDNAs using real-time PCR were listed in Key resources table.
Proliferation assay Two thousand myoblasts or mesenchymal progenitors were seeded in 96-well plate and were cultured in DMEM-LG containing 10% FCS, and penicillin-streptomycin. Proliferation assays were performed using Real-time Glo (Promega) and the luminescence were measured by a Glomax Microplate Reader.
Single myofiber culture with supernatant experiments from mesenchymal progenitors Pdgfra + CD31 À CD45 À cells from wild-type or Thbs1-KO mice were cultured on a collagen-coated dish (Iwaki, #4020-010) for 72 h in DMEM-LG containing 10% FCS, and penicillin-streptomycin. Freshly isolated EDL myofibers from female C57BL/6 mice were maintained in a 1:1 mixture of culture supernatant and fresh culture medium (DMEM-LG containing 10% FCS, and penicillin-streptomycin) that was not refreshed during the cultivation. The myofibers were fixed with 2% PFA at 48 and 72 h after the cultivation. Finally, the number of Pax7-positive cells was counted.

QUANTIFICATION AND STATISTICAL ANALYSIS
Values are expressed as means ± SD. Statistical comparison between two groups was performed by two-sided unpaired Student's t test. For comparison of more than two groups, one-way ANOVA and Tukey-Kramer test (excluding Figure 5G) or Dunnett test (Figure 5G) were used. A p value less than 0.05 was considered to be statistically significant.